What is the Difference between Earthing, Grounding and Bonding?
There is unusual confusion to understand the basic concept and main difference among grounding, earthing and bonding even some professionals interchanged the word for earthing, grounding and bonding such as earthing bond, bonding ground etc. In addition, electrical bonding is a totally different thing other than grounding and earthing.

To the point, Grounding and Earthing is the same concept expressed by different terms used for them. There is a small difference between earthing and grounding which we will discuss in detail

Case study: radiated interference to industrial controller
https://www.edn.com/case-study-radiated-interference-to-industrial-controller/?utm_content=buffer03cba&utm_medium=social&utm_source=edn_facebook&utm_campaign=buffer

As an EMC consultant, I seem to be running into more and more issues with ESD and radiated susceptibility. I believe this is due to the fact noise margins are gradually being reduced as supply voltages move from 5 to 3.3 to 1.8 to 1.2 volts. In addition, IC chips are scaling down in size, and quite frankly, designers still don’t understand basic EMC design principles, as I wrote up recently in an editorial for Interference Technology’s 2014 Test & Design Guide

Generally, the first thing I like to do is to sniff around with a near field probe and current probe to get a feel for any radiated emission issues. Finding nothing major, the project engineer demonstrated how he could affect the controller using just a Family Radio Service (FRS) walkie talkie from about 10 feet away. I recently measured a typical FRS radio at a 1m test distance and it read about 2V/m. Using Equation 1, at 3m (about 10 feet), we’re talking just a 1.3 V/m field strength, where I’m assuming the actual power output from the FRS radio is 0.25W, the antenna gain is 0.7 and the distance is 3m.

We actually performed most of the testing using that FRS radio. Initially, though, the resolution using the radio was too coarse, so a near field probe was connected to an RF generator, tuning it to one of the failing frequency bands (Reference 6). By probing around, we narrowed the issue down to one of several cables running through a mechanical arm on the machine.

A shielded box with several cables running through grommets. Penetrating a shield with a cable without terminating the shield allows RF interference into the enclosure.

we discovered the designer had failed to connect the cable shield! Once the cable shield was bonded to the chassis structure at both ends, the controller was completely immune to RF signals.

It’s my experience that many designers seem unsure how and where to connect cable shields. I’m simplifying somewhat, but connecting the shield at one end provides a good E-field shield. Connecting it at both ends (to the same structure) provides a good H-field shield. Most digital circuitry relies on low impedance, low voltage switched currents. Therefore, it’s more important to shield for the resulting H-fields. On the other hand, things like switch mode power supplies utilize high impedance with switched high voltages and so E-field shielding is a practical solution. Additionally, connecting each end of the shield to two differing potentials – for example, one end to digital return and one end to chassis – can introduce a potential difference which can inject high frequency switching noise into the signal wires.

There’s an additional point to be made regarding cable shields and that is the type and quality of the shielding material. Some less expensive cables use loosely formed shielding, with distributed gaps along the length. These should be avoided, due to poor shielding effectiveness. More expensive cables have a tighter weave on the shield with correspondingly better shielding performance.

In conclusion, it turns out that most of the client projects in which I’ve been involved that fail one, or more, EMI tests are due to basic design issues, such as poor routing of clock traces, penetration of I/O cables through shielded enclosures, and poor termination of cable shields. For more on shielding and bonding, check the references.

The ground wire is not usually at exactly ground potential due the impedance (usuallt around same as neutral impedance) and the leakage current that gets coupled to it (wire capacitance, equipment Y capacitors in RFI filter etc.).
And in addition there are inductively coupled voltage and possibly some RFI coupled to ground wire.

This KnowledgeBase provides general guidelines for installing National Instruments test and measurement equipment that require a connection to the facility grounding system for the purpose of enhancing electromagnetic compatibility (EMC) performance in accordance with the product documentation.

Of the different choices that are available for grounding a shield braid that encloses a differential pair of signal wires, please consider that the shield braid be grounded only at the signal source, at the input end, and not at the output end.

As a first thought, and as something that is often advocated, grounding a shield at both ends may result in severe ground loop currents which could adversely impact EMI and isolation properties.

As a second thought, with the shield grounded only at the output end
the interground interference signal, Enoise, can induce a differential noise signal between the two outputs E1 and E2 that feed the differential amplifier

As a third thought, grounding the shield only at the input end averts both the ground loop problem and the time constant mismatch problem.

Enoise as a common mode signal so that no differential voltage is created between E1 and E2. The A2 differential amplifier is thereby protected from Enoise.

The term “ground” should be used more carefully. In our equipment, we distinguish between ground and DC common for good reason. Ground refers to chassis or safety ground to take errant AC voltage to ground instead of through an operator’s body.

Recently, I discovered that more than one of the major industrial LCD manufactures tie the metal frame (chassis) of the LCD to DC common. When asked about this, they replied, “To reduce EMI.” This may be fine to make the display pass emission standards, but it is just plain wrong from an electronic design point of view. Industrial equipment (the intended market) is often housed in metal enclosures tied to AC ground.

This one disturbed the nonvolitile Flash memory, causing the CPU configuration data to be corrupted. The operator was rather dismayed to encounter a “keyboard error” message on an embedded system that had no keyboard. The most amusing part of the message was the line that said to “Press F1 to continue.”

Lesson for the day: Ground and common are not the same, and never the twain shall meet, unless made by monkeys with EMI in their heads.